
Probable MCP server
Which servers in the official MCP registry are Probable's own, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.
read 2026-10-26
- Registry namemedia.probable/newsverified in the MCP registry
The registry name sits under probable.media. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.
- Stateanswerschecked 2026-10-26
- Tools14
- Server calls itselfprobable-media
- Server version1.7.0
- Protocol version2025-06-18
On 2026-10-26 the server answered initialize and tools/list without a sign-in. First answer in 92 ms.
- Its tools do every use case we checked: Pair today's story with a different event from the same topic; Skim five sourced stories in a short daily brief.
- Checked on 1 of the 7 days it takes to judge reliability.
Once connected
- For "Pair today's story with a different event from the same topic", the likely first call is read_latest_edition.
- For "Skim five sourced stories in a short daily brief", the likely first call is get_brief.
- create_watch, update_watch change data without a sign-in: anyone with the address can call them.
- Parameters are untyped or undescribed in 7 of 14 tools: check arguments before calling.
Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.
- INITE MCP scoreafter a weeka score needs 7 daily checks: 1 so far
- What an agent can do with it2 of 2 use cases
- Answeredmeasuring1 of the 7 daily checks it takes to print a share
- Median first answer474 ms
- Space in an agent's context≈ 2,522 tokensthe definitions of all its tools (14), which an agent carries on every turn
- Tool definitionsA · 92 of 100
- Typed results0 of 14 toolsdeclared outputSchema: results come typed, read once
- Parameters without a description: in 7 of 14
What an agent can do with it
- not sure · Place a headline among related reporting in the observed corpus
- can do · Pair today's story with a different event from the same topic
- can do · Skim five sourced stories in a short daily brief
Each of the product's use cases, put to Jev against this server's tool list: can an agent do it with these tools?
Worth a look
- Tools that change data, on a server that answers without a sign-in: create_watch, update_watch
A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (30%), doing the product's job (25%), definitions (20%), context (10%) and use (15%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →
read_latest_editionRead the latest source-grounded Probable edition. Every replacement is a different event observed in the same scan and category with a higher editorial comparison index; that index is not a measured occurrence rate.
get_briefUse for a current news brief or catch-up job. Return a five-story cited brief diversified across category, event class, and publisher by default. Use chronological or all for stable cursor paging. A balanced resumeCursor continues as a…
get_contextUse when a user supplies a headline or a source URL already observed by Probable and wants same-class context. Optional referenceContext adds explicit country, magnitude, industry or aviation scope for primary-source population…
list_editionsList every stored edition with public metadata, free/paid status, exact prices, and the URLs that can unlock paid content.
read_edition_by_dateRead a stored edition inside the seven-day public window. Older editions return an access boundary and links; this tool never bypasses membership or payment.
export_editionExport an edition as source-grounded Markdown or CSV. Requires member authorization with exports:read and an active Founding Pro entitlement.
list_watchesList the member's saved monitors and recent matches. Requires member authorization with watches:read and an active Founding Pro entitlement.
create_watchCreate a Founding Pro saved watch. Requires watches:write; repeat the same Idempotency-Key to safely replay the same request.
update_watchUpdate fields on one owned Founding Pro watch. Requires watches:write and an Idempotency-Key.
acknowledge_watchMark all currently unread matches on one owned watch as read. Requires watches:write and an Idempotency-Key.
discover_eventsUse to rank cited events in the latest completed edition by probable, rare, or overweighted mode. Keep research priors separate from measured current-edition counts. Do not describe the result as a forecast or a probability that an event…
score_headlineClassify one supplied headline and inspect coverage recurrence. Optional referenceContext adds explicit country, magnitude, industry or aviation scope. When sufficiently scoped, occurrenceReference separately supplies primary-source…
list_classesReturn the versioned category and event-class catalog used by score_headline. The editorial index is a legacy comparison weight, not an empirical probability.
get_membershipReturn Free and Founding Pro terms and available Stripe-hosted checkout URLs. Calling this tool does not open checkout or claim payment success.
How to connect
Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:
{
"mcpServers": {
"news": {
"type": "http",
"url": "https://probable.media/mcp"
}
}
}Add it
Claude Code, in a terminal
claude mcp add --transport http probable https://probable.media/mcpclaude.ai, ChatGPT and other clients: add https://probable.media/mcp as a custom connector in the settings.
OpenAPI spec
36 paths, read 2026-10-26. Open the spec ↗
Each line is something a program found when it went looking: the registry entry, the server's own answer, a spec that parses. A missing line means we did not find it, not that it does not exist.
Cookies
Before we count this visit
Analytics and ad measurement tell us which pages and which of our ads do their job: Google Analytics and Google Ads, the Meta pixel, Yandex Metrica, and one first-party id of our own. Nothing is stored until you answer, and declining changes nothing about the site.